# Out-of-Sample Prediction: Train ≤2006, Test 2007–2012

| Dep. Variable | Model | Train N | Test N | Train R² | AUC | Recall | Precision | F1 | Actual | Predicted |
|:---|:---|---:|---:|---:|---:|---:|---:|---:|---:|---:|
| banking_crisis_onset | EW only | 2492 | 989 | 0.0111 | 0.458 | 0.00 | 0.00 | 0.00 | 25 | 16 |
| banking_crisis_onset | EW + Z | 2492 | 989 | 0.0122 | 0.517 | 0.00 | 0.00 | 0.00 | 25 | 18 |
| any_crisis_onset | EW only | 2492 | 989 | 0.0337 | 0.516 | 0.05 | 0.14 | 0.07 | 41 | 14 |
| any_crisis_onset | EW + Z | 2492 | 989 | 0.0353 | 0.514 | 0.05 | 0.11 | 0.07 | 41 | 18 |
| ca_reversal | EW only | 2492 | 989 | 0.0267 | 0.606 | 0.06 | 0.67 | 0.11 | 228 | 21 |
| ca_reversal | EW + Z | 2492 | 989 | 0.0470 | 0.625 | 0.27 | 0.42 | 0.33 | 228 | 146 |

*Threshold = training-set event base rate. AUC computed from predicted scores (Xβ without fixed effects).*